This paper uses a regression discontinuity design to investigate the effects of a school referral policy on the academic achievement, well-being and social relations of a subgroup of bilingual pupils who do not speak the host-country language at home. Since 2006, all bilingual school starters living in the second largest city in Denmark are required to take a Danish language test shortly before school start. Children whose scores fall below a pre-determined threshold are referred to a school with at most 20% pupils with inadequate language proficiency per school. Moreover, if the school is located outside the school district, the local authorities provide the pupils with free bus services between home and school. Comparing pupils below and above the threshold,
our very preliminary results show a significant but very small negative effect on the probability to attend the after-school program of the school.